Struct PerConfig

pub struct PerConfig {
    pub alpha: f32,
    pub beta_0: f32,
    pub beta_final: f32,
    pub n_opts_final: usize,
    pub normalize: WeightNormalizer,
}

Configuration for Prioritized Experience Replay (PER).

This structure defines the parameters for prioritized sampling in the replay buffer. It controls how transitions are sampled based on their importance and how importance weights are calculated and normalized.

Fields

• alpha - Controls the degree of prioritization (0 = uniform sampling)
• beta_0 - Initial value for importance sampling weights
• beta_final - Final value for importance sampling weights
• n_opts_final - Number of optimization steps to reach beta_final
• normalize - Method for normalizing importance weights

Examples

use border_core::generic_replay_buffer::{PerConfig, WeightNormalizer};

let config = PerConfig::default()
    .alpha(0.6)
    .beta_0(0.4)
    .beta_final(1.0)
    .n_opts_final(500_000)
    .normalize(WeightNormalizer::All);

Fields

alpha: f32

Exponent for prioritization. Higher values increase the bias towards high-priority transitions. A value of 0 results in uniform sampling.

beta_0: f32

Initial value of the importance sampling exponent. Lower values reduce the impact of importance sampling weights.

beta_final: f32

Final value of the importance sampling exponent. Typically set to 1.0 to fully compensate for the non-uniform sampling.

n_opts_final: usize

Number of optimization steps after which beta reaches its final value. This allows for a gradual increase in the impact of importance sampling.

normalize: WeightNormalizer

Method for normalizing importance sampling weights. Controls how the weights are scaled to prevent numerical instability.

Implementations

impl PerConfig

pub fn alpha(self, alpha: f32) -> Self

Sets the prioritization exponent alpha.

Arguments

• alpha - The new value for the prioritization exponent

Returns

The modified configuration

pub fn beta_0(self, beta_0: f32) -> Self

Sets the initial importance sampling exponent beta_0.

Arguments

• beta_0 - The new initial value for the importance sampling exponent

Returns

The modified configuration

pub fn beta_final(self, beta_final: f32) -> Self

Sets the final importance sampling exponent beta_final.

Arguments

• beta_final - The new final value for the importance sampling exponent

Returns

The modified configuration

pub fn n_opts_final(self, n_opts_final: usize) -> Self

Sets the number of optimization steps to reach the final beta value.

Arguments

• n_opts_final - The new number of optimization steps

Returns

The modified configuration

pub fn normalize(self, normalize: WeightNormalizer) -> Self

Sets the method for normalizing importance weights.

Arguments

• normalize - The new normalization method

Returns

The modified configuration

Trait Implementations

impl Clone for PerConfig

fn clone(&self) -> PerConfig

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for PerConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for PerConfig

fn default() -> Self

Creates a default PER configuration with commonly used values:

• alpha = 0.6 (moderate prioritization)
• beta_0 = 0.4 (initial importance sampling)
• beta_final = 1.0 (full compensation)
• n_opts_final = 500_000 (gradual increase)
• normalize = All (normalize all weights)

impl<'de> Deserialize<'de> for PerConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for PerConfig

fn eq(&self, other: &PerConfig) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for PerConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for PerConfig